UX WORK › CASE STUDY
One Account,
One Profile.
UX WORK › CASE STUDY
One Account,
One Profile.
Using identity infrastructure to unlock candidate data for 100M+ users across Indeed and Glassdoor.
Role
UX Lead for Auth Experience
Timeline
April 2025 – February 2026
Scope
End-to-end authentication UX, consent flows, cross-platform identity Team UX, Product, Engineering, Legal, Brand, Glassdoor Platform
Team
3 UX, 2 Product, 2 Content, 10+ Engineering, Legal, Brand
TL;DR
The problem
Glassdoor job seekers were getting worse matching outcomes than Indeed users, lower placement rates, lower application quality, because Indeed knew far less about them. No shared identity layer meant profile data couldn't cross platforms.
What I did
Led end-to-end UX for a trust-first account connection experience across 5 auth states and 3 sign-in methods, while filling the cross-functional coordination gap.
The shift
Reframed the design problem from "reduce friction" to "build trust first".
The result
Resume coverage went from 33% → 70%. Sourcing opt-in from 29% → 54%. 54,000 recruiter outreaches in the first month.
OUTCOMES
70%
users with a resume
↑ from 33%
54%
resume + sourcing opt-in
↑ from 29%
54K
recruiter outreaches, month one
US alone
20K
positive connections, month one
sourcing channel now open
THE PROBLEM
Parallel universes, invisible users.
Glassdoor job seekers were getting worse outcomes than Indeed users, and the root cause was structural, not behavioral.
Indeed supplies job search and recommendations to Glassdoor. But the two platforms operated as parallel universes. A job seeker could have a complete Indeed profile (resume, salary expectations, sourcing consent) and none of that would exist on their Glassdoor side. The result: lower placement outcomes and application quality scores for Glassdoor users, because Indeed simply knew less about them.
OAOP (One Account, One Profile) was the initiative to close that gap by establishing Indeed as the identity provider for Glassdoor, so profile data, resume, and preferences could be unified across both platforms. The goal was better outcomes for job seekers and employers who were being failed by a structural gap they couldn't see.
MY ROLE
Auth was the brief. The gap was the opportunity.
As the sole UX designer on Indeed’s Authentication Experience team, I owned the end-to-end auth flow for OAOP, from consent through post-connection, across SMS/Email OTP, passkeys, social sign-on, and Google One Tap.
OAOP was a large, cross-org initiative spanning 20 work streams across Indeed and Glassdoor. As work progressed, a coordination gap emerged across UX, Engineering, Product, Legal, and Content, with no single owner connecting it all. I stepped into that role, keeping the full user journey in view and driving decisions across teams with competing priorities.
I also pulled content into the process early, establishing a dedicated partnership and shared workflow to ensure alignment across both product and messaging.
DISCOVERY
The real friction wasn't in the flow.
THE INSIGHT
Glassdoor users expect anonymity; it's where people go to research companies and leave candid reviews. Asking them to connect their identity to Indeed felt like the terms of an unspoken agreement were changing.
And there was a hard constraint that sharpened this further: for users with existing Glassdoor accounts, we could not require connection. A genuine opt-out had to exist. Which meant if the experience felt coercive, users would simply decline and the entire initiative would fail quietly.
THE DESIGN
Consent as a moment, not as a checkbox.
CONSENT ARCHITECTURE
Every user should understand what they're sharing, with whom, and why, before they agree to anything. Rather than burying data-sharing language in legal copy, I translated the technical reality into plain language benefit. The consent architecture satisfied Legal's requirements without letting compliance copy overwhelm the clarity users needed to say yes.
GOOGLE ONE TAP
Speed without sacrifice; One Tap is fast, but when users can connect accounts in a single tap, the moment passes before they've processed what happened. The final design preserved the speed of One Tap while surfacing enough context: what's being connected, what's being shared; so users felt informed rather than automatically enrolled.
AUTHENTICATION STATE SYSTEM
Users arrived from very different starting points. I mapped the full state matrix so every path got the right flow, not just the common one.
State 01 surfaced the most complex design moment in the project: users with a resume on both platforms had to choose which to keep, with the other deleted along with its parsed data.
CATCHING PROBLEMS EARLY
Two moments stand out, both required the confidence to push back on work that was already in motion.
The first was Joint ToS: I caught content inconsistencies between the Indeed and Glassdoor terms of service surfaces before implementation and drove resolution with Legal. The language users were being asked to agree to wasn't aligned across both platforms, a problem that would have been significantly harder to unwind after launch.
The second surfaced post-implementation. When users clicked the "Learn More" link on web, the login popup was hidden when they returned to the original tab, because the popup ran as a separate browser window. Users who tried to get more information before consenting lost their place entirely.
I proposed resolving this by implementing "Learn More" as a contained screen within the popup itself, a separate screen users could navigate to and back from without ever leaving the flow. This avoided the redirect entirely, kept users oriented, and was the only approach Legal approved as a permanent solution. The tradeoff was engineering complexity against the MVP timeline, and limited screen real estate while Legal was still drafting the required content, both of which I documented explicitly as the decision moved forward.
RESULTS
A sourcing channel that actually worked.
The patterns established through OAOP became the foundation for what the platform can do next, OTP content guidelines now used by other teams, and an OAuth audit that connected to white-label scaling work globally.
In January 2026 alone (the first month at full ramp) Glassdoor generated 54,000 recruiter outreaches and 20,000 positive connections. A sourcing channel that had been functionally closed before.
EXTENDING INTO SUPPORT INFRASTRUCTURE
Designing OAOP meant thinking beyond the auth flow, into what happens when users need help, and how support teams would understand the new connected account state.
As part of the rollout, the Job Seeker Support (JSS) team conducted discovery to understand what they'd need to support OAOP going forward. One key requirements emerged: support agents needed a way to see whether an account had been successfully connected to Glassdoor.
The simplest solution was surfacing connection status directly in the existing Account Admin lookup tool. But I used the opportunity to make a second improvement alongside it: the original tool returned results as a single line of inline text: User ID, email, phone number, and date all strung together, which made scanning across multiple results slow. I restructured the results into a proper table, with each data point in its own labelled column. The "Connected to Glassdoor" field was added as part of that same update.
The before/after shows both changes together: results that were a line of text became a scannable table, and support agents gained a clear, timestamped connection status at a glance.
TAKEAWAY
Infrastructure work with a human problem at its center.
The authentication layer was the mechanism. The goal was always something bigger: richer profiles, better matches, a sourcing channel that actually worked.
Getting there meant catching risks early, building coordination the work required but no one had assigned, and designing consent as the moment it actually is, not a formality, but the point where users decide whether to trust you.
The profiles are richer now. The sourcing channel is open. That's what the work was for.